The Smad6-Histone Deacetylase 3 Complex Silences the Transcriptional Activity of the Glucocorticoid Receptor

Glucocorticoids play pivotal roles in the maintenance of homeostasis but, when dysregulated, may also have deleterious effects. Smad6, one of the transforming growth factor β (TGFβ) family downstream transcription factors, interacts with the N-terminal domain of the glucocorticoid receptor (GR) through its Mad homology 2 domain and suppresses GR-mediated transcriptional activity in vitro. Adenovirus-mediated Smad6 overexpression inhibits glucocorticoid action in rat liver in vivo, preventing dexamethasone-induced elevation of blood glucose levels and hepatic mRNA expression of phosphoenolpyruvate carboxykinase, a well known rate-limiting enzyme of liver gluconeogenesis. Smad6 suppresses GR-induced transactivation by attracting histone deacetylase 3 to DNA-bound GR and by antagonizing acetylation of histone H3 and H4 induced by p160 histone acetyltransferase. These results indicate that Smad6 regulates glucocorticoid actions as a corepressor of the GR. From our results and known cross-talks between glucocorticoids and TGFβ family molecules, it appears that the anti-glucocorticoid actions of Smad6 may contribute to the neuroprotective, anticatabolic and pro-wound healing properties of the TGFβ family of proteins.

Glucocorticoids play crucial roles in the regulation of basal and stress-related homeostasis (1). They are necessary for maintenance of many important biologic activities, such as homeostasis of intermediary metabolism, central nervous and cardiovascular system functions, and the immune/inflammatory reaction (1). Glucocorticoids also act as potent immunosuppressive and anti-inflammatory agents at "pharmacologic" doses, properties that make them irreplaceable therapeutic means for many inflammatory, autoimmune, allergic, and lymphoproliferative diseases (2). At high levels and over a long duration, glucocorticoids have neurotoxic, catabolic, and anti-wound healing properties as well as promoting gluconeogenesis, adipogenesis, and a shift of the T helper 1 to T helper 2 balance toward T helper 2 predominance (3,4). Many extracellular and intracellular factors influence the actions of glucocorticoids at the level of their target tissues (5). Some of these are physiologically important, whereas others may also be associated with pathologic processes (5,6).
The actions of glucocorticoids are mediated by the ubiquitous intracellular glucocorticoid receptor (GR), 2 which functions as a hormoneactivated transcription factor of glucocorticoid target genes (3,7). The GR consists of three domains, the N-terminal or "immunogenic" domain, the central, DNA-binding domain (DBD), and the C-terminal, ligand-binding domain. The functions of the latter two domains have been studied extensively, whereas those of the immunogenic domain are less well known (7). In the unliganded state, the GR is located primarily in the cytoplasm (7). After binding to its agonist ligand, the GR undergoes conformational changes and translocates into the nucleus. Ligand-activated GR then binds to the glucocorticoid response elements (GREs) as a dimer and attracts several so-called coactivators and chromatin-remodeling factors to the promoter region through its two transactivation domains, activation function (AF)-1 and AF-2 (7,8). Among them, the p160 type histone acetyltransferase coactivators play an essential role in GR-induced transcriptional activity, being attracted to the promoter region in an early phase of transcriptional activation and facilitating access of other transcription-related molecules on the chromatin through acetylation of lysine residues located in several histone tails, such as those of histone H3 and H4 (8 -12). In contrast, corepressors, such as the nuclear receptor corepressors and the silencing mediator for retinoid and thyroid hormone receptor, and associated histone deacetylases cause deacetylation of histones, silencing gene transcription by preventing access of cis-acting molecules to the promoter region (8).
Members of the Smad family of proteins transduce signals of transforming growth factor ␤ (TGF␤) superfamily members, such as TGF␤, activin, and bone morphogenetic proteins (BMPs), by associating with the cytoplasmic side of the type I cell surface receptors of these hormones (13,14). Nine distinct vertebrate Smad family members have been identified, and they have been classified into three groups: receptor-regulated Smads (R-Smads), such as Smad1, -2, -3, -5, and -8; a common partner Smad (Co-Smad), Smad4; and inhibitory Smads (I-Smads) like Smad6 and Smad7 (13).
All Smads have two characteristic domains, the Mad homology domains 1 and 2 (MH1 and -2), in their N-terminal and C-terminal portions, respectively, separated by a proline-rich linker region (13). The MH1 domain of R-and Co-Smads is important for complex formation with other Smads, transcriptional activation and repression, and interaction with other transcription factors and target DNA sequences (13). The MH2 domain of R-Smads mediates their interaction with cell surface receptors (13,15), whereas the highly conserved MH2 domain of I-Smads interacts with type I receptors and is sufficient for their inhibitory activity (14).
Upon binding of TGF␤, activin, or BMP to their receptors, cytoplasmic R-Smads are phosphorylated by the receptor kinases, translocate into the nucleus, and stimulate the transcriptional activity of TGF␤-, activin-, or BMP-responsive genes by binding to their response elements located in their promoter regions as a heterotrimer with Co-Smad (13). I-Smads, such as Smad6 and Smad7, act as inhibitory molecules in the TGF␤ family signaling by forming stable associations with activated type I receptors, which prevent the phosphorylation of R-Smads (13). Smad6 also competes with Smad4 in the heteromeric complex formation induced by activated Smad1 (16). In addition, I-Smads directly suppress the transcriptional activity of TGF␤ family signaling by binding to promoter DNA and attracting histone deacetylases and/or the C-terminal binding protein (17)(18)(19). Since I-Smads are produced in response to activation of TGF␤ family signaling (20), they literally function in the negative feedback regulation of the Smad signaling pathways. Smad6 preferably inhibits BMP signaling, whereas Smad7 is a more general inhibitor, repressing TGF␤ and activin signaling in addition to that of BMP (14).
In this study, we found that Smad6 interacts with the GR and suppresses the latter's transcriptional activity by attracting histone deacetylase HDAC3. HDAC3 antagonized histone acetylation induced by p160 type histone acetyltransferase coactivators. This inhibitory effect of Smad6 was present in vivo, suppressing glucocorticoid-stimulated gluconeogenesis in the rat liver. It is likely that Smad6 functions as a target tissue regulator of glucocorticoid action.
Yeast Two-hybrid Screening and Assay-The yeast two-hybrid screening was performed using GR-(263-419) as bait in the human Jurkat cell cDNA library with the LexA system (Clontech). For a yeast two-hybrid assay, yeast strain EGY48 (Clontech) was transformed with pOP8-LacZ and the indicated pLexA-and pB42AD-based plasmids. ␤-Galactosidase activity was then measured in the cell suspension as previously described (26). The ␤-galactosidase activity was normalized for A 600 nm . -Fold induction was calculated by the ratio of adjusted ␤-galactosidase values of transformed cells cultured in the presence of galactose/raffinose versus those in the medium containing glucose.
Cell Cultures and Transfection-Human colon carcinoma HCT116 and uterine cervical carcinoma HeLa cells were purchased from the American Type Culture Collection (Manassas, VA) and were maintained in McCoy's 5A or Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units of penicillin, and 50 g/ml streptomycin. HCT116/MMTV cells, which were stably transformed with pMAM-neo-Luc that has the full-length MMTV promoter upstream of the luciferase gene, were maintained in the McCoy's medium containing 0.2 mg/ml neomycin and the same supplements. Rat hepatoma HTC cells were described previously (21). HCT116 and HCT116/MMTV cells do not contain functional GR, whereas HeLa and HTC cells express fully active GR.
HCT116 and HCT116/MMTV cells were transfected as previously described (21). For the experiments using pMMTV-Luc or other reporter genes for indicated nuclear receptors and transcription factors, different amounts of Smad6-or Smad7-related plasmids were cotransfected with 0.5 g/well of the indicated nuclear receptor-or transcription factor-expressing plasmid, 1.5 g/well of luciferase-expressing reporter plasmid, and 0.5 g/well of pSV40-␤-Gal. For stimulation of p53, CREB, or NF-B transcriptional activity, 0.5 g/well of pRc/CMV-hp53, RSV-PKA, or pRSV-RelA was cotransfected as well. Empty vectors were used to maintain the same amounts of transfected DNA. 10 Ϫ6 M dexamethasone or progesterone, or 10 Ϫ8 M aldosterone, dehydrotestosterone, or estradiol was added to the medium after 24 h of transfection. The indicated concentrations of tricostatin A (TSA) (Sigma), the his-tone deacetylase inhibitor, were also added to the medium at the same time. The cells were harvested after an additional 24 h, and luciferase and ␤-galactosidase assays were performed as previously described (24).
Introduction of Smad6 Small Interfering RNAs (siRNAs) into HTC Cells, the Tyrosine Aminotransferase (TAT) Assay, and the Real Time PCR-The rat Smad6 siRNA (5Ј-GCCACUGGAUCUGUCCGAUd-TdT-3Ј), which targets nucleotides 945-964 of the coding region (Gen-Bank TM accession number XM_345947), was produced by Qiagen (Valencia, CA). This sequence portion has the complete match to corresponding parts of the mouse and the human Smad6 sequences. The negative control siRNA (5Ј-UUCUCCGAACGUGUCACGUdTdT-3Ј) was also purchased from Qiagen.
HTC cells were transfected with siRNAs by using the Nucleofector system (Amaxa GmbH, Cologne, Germany) with nearly 80% transfection efficiency, as previously described (21). Twenty-four hours after plating the cells in the 24-well plates, they were stimulated with 10 Ϫ6 M dexamethasone. After an additional 24 h of incubation, cell lysates for the tyrosine aminotransferase (TAT) assay and total RNA for the real time PCR were harvested. TAT assays were performed as previously reported (21).
The reverse transcription reaction was carried out as previously described (27). To detect mRNA levels of rat Smad6 and control rat acidic ribosomal phosphoproteinP0(RPLP0),primerpairs(Smad6,forward(5Ј-GAAGTCGT-GTGGTCCCTGATC-3Ј) and reverse (5Ј-CTCGCAGTCACTCTCAG-3Ј); RPLP0, forward (5Ј-GACATGCTGCTGGCCAATAAG-3Ј) and reverse (5Ј-CAACATGTTCAGCAGTGTG-3Ј)) were used (21). The real time PCR was performed in triplicate using the SYBR Green PCR Master Mix (Applied Biosystems) in an ABI PRIZM 7700 SDS light cycler (Applied Biosystems), as previously described (27). Obtained CT (threshold cycle) values of Smad6 were normalized for those of RPLP0, and their relative mRNA expression was demonstrated as -fold induction to the base line. The dissociation curves of the used primer pairs showed a single peak, and samples after PCRs had a single expected DNA band in an agarose gel analysis (data not shown).
FLAG-Smad6-expressing Adenovirus, Injection into Rats, and Detection of the Phosphoenolpyruvate Carboxykinase (PEPCK) and TAT Gene mRNAs-The following animal study was approved by the NICHD Animal Care and Use Committee (protocol number ASP04-008). FLAGtagged Smad6-expressing and control LacZ-expressing adenoviruses were described previously (28). 1 ϫ 10 10 colony-forming units of these adenoviruses were injected into the peritoneal cavity of 175-200-g Sprague-Dawley rats, and 1 mg/kg dexamethasone was injected intramuscularly after 24 h. After an additional 24 h, blood glucose levels were examined in total blood obtained by tail cut using a OneTouch monitor (LifeScan, Milpitas, CA). The rats were then sacrificed with CO 2 , and the livers were removed and stored at Ϫ70°C until further use. Total RNA and whole homogenate of the liver were subsequently purified using the RNAeasy Midi kit (Quiagen) and homogenation/centrifugation at 200 ϫ g for 10 min in the buffer containing 50 mM Tris-HCl (pH 7.4), 50 mM NaCl, 0.2% Nonidet P-40, and Complete TM tablets (1 tablet/50 ml), respectively. Levels of the PEPCK, TAT, and RPLP0 mRNA were then determined with the real time PCR reaction in quadruplicate using the SYBR Green PCR Master Mix, as described above. Primer pairs for detecting mRNAs of the hepatic PEPCK and TAT were as follows: PEPCK, forward (5Ј-CAGGCTGGCTAAGGAGGAAG-3Ј) and reverse (5Ј-CATCACTTGTCTCAGCCAC-3Ј); TAT, forward (5Ј-GTCGCTTCTTACTACCAC-3Ј) and reverse (5Ј-CAGGCAGGAG-ATTGTAGAG-3Ј) (21,27). Whole homogenate of the liver was run on 8% SDS-polyacrylamide gels, and levels of GR and FLAG-Smad6 were examined in Western blots using anti-GR (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and anti-FLAG (M2) (Sigma) antibodies, respectively.
Chromatin Immunoprecipitation (ChIP) Assay-The ChIP assay was performed in HCT116/MMTV and HTC cells, which have the genomically integrated MMTV-luciferase gene (HCT116/MMTV) and endogenous glucocorticoid-responsive TAT gene (HTC), respectively, using a chromatin immunoprecipitation kit (Upstate, Charlottesville, VA) with minor modifications as previously described (27,29). HCT116/MMTV cells were transfected with FLAG-Smad6-and/or HA-HDAC3-expressing plasmids together with pRShGR␣ using Lipofectin. HTC cells were transfected with FLAG-Smad6-or GRIP1 (GR-interacting protein 1)-expressing plasmid using the Nucleofector system. Both cells were exposed to either 10 Ϫ6 M dexamethasone or vehicle overnight. They were then fixed, DNA and bound proteins were cross-linked, and ChIP assays were performed by co-precipitating the DNA-protein complexes with anti-GR␣ (Affinity Bioreagents, Golden, CO), anti-FLAG (M2), anti-HA (Santa Cruz Biotechnology), anti-GRIP1 antibodies (Santa Cruz Biotechnology), or rabbit control IgG (Santa Cruz Biotechnology). Antibodies reacting to the acetylated histone H3 (Lys 14 ) and H4 were purchased from Upstate. The promoter region Ϫ219 to Ϫ47 of the MMTV long terminal repeat (fragment size 173 bp), which contains two functional GREs, was amplified from the prepared DNA samples using a primer pair: 5Ј-AACCTTGCGGTTCCCAG-3Ј and 5Ј-GCATTTA-CATAAGATTTGG-3Ј (29). Tandem endogenous GREs of the rat TAT promoter, which are located ϳ2,500 bp upstream of its transcription initiation site, were amplified by a primer pair: 5Ј-TCTTCTCAGTGT-TCTCTATCAC-3Ј and 5Ј-CAGAAACCGACAGGCGACTACG-3Ј (fragment size 220 bp), as described previously (27). Amplified products were then run on a 3% agarose gel, and visualized DNA bands were photographed.
The real time PCR was directly performed using the primer pair that detects TAT GREs for some of the ChIP samples using the SYBR Green PCR Master Mix (Applied Biosystems) as described above. Obtained CT values of ChIP samples were normalized for those of corresponding inputs, and their relative precipitations were demonstrated as -fold precipitation of the base line.
Statistical Analyses-Statistical analysis was carried out by analysis of variance, followed by Student's t test with Bonferroni correction for multiple comparisons or unpaired t test with the two-tailed p value.

Smad6 Interacts with GR-(263-419) through Its C-terminal Portion-
The N-terminal domain of the human GR consisting of 420 amino acids accounts for over a half of the entire molecule (7). Although it contains the AF-1 domain at amino acid positions 77-261, through which the GR communicates with components of the transcriptional machinery (7), functions of the rest of the N-terminal domain are yet unknown. Thus, we performed a yeast two-hybrid screening assay using as bait a GR fragment spanning amino acids 263-419, located between the AF-1 domain and the DBD in a Jurkat cDNA library. Among over 85 independent interactors, we found two independent clones containing the C-terminal portions of the human Smad6 coding sequence (data not shown). In a reconstituted yeast two-hybrid assay, GR-(263-419), expressed as a fusion with the B42 activation domain (AD), interacted with the full-length and C-terminal portion (amino acids 331-496) of Smad6 expressed as a fusion with LexA-DBD, but not with a Smad6 fragment containing amino acids 1-330 (Fig. 1A). These results indicate that GR-(263-419) interacts with the C-terminal portion of Smad6, which corresponds to the MH2 domain of this molecule. Since we have also found that the WD repeat proteins, the guanine nucleotide-binding protein ␤ (G␤), and Rack1 (receptor for activated protein kinase C 1) interact with GR (263-419) in addition to Smad6 (21), we examined whether Smad6 and G␤2 interact with different portions of GR-(263-419) in a yeast two-hybrid assay. Several plasmids expressing three different portions of GR-(263-419) expressed as fusions with the LexA-DBD were constructed, and their binding activity to B42 AD-fused Smad6-(319 -496) and G␤2-(55-226) was tested. The latter corresponds to blades 1-4 of G␤2 and demonstrated strong interaction with GR-(263-419) in our previous examination (21) (Fig. 1B). Smad6 interacted with GR-(263-419) strongly but not with the other shorter fragments of this portion of the GR. In contrast, G␤2 bound to GR-(263-319) and GR-(263-367) as well as to GR-(263-419), but not with GR-(319 -367) and GR-(367-427). These results indicate that the entire GR-(263-419) is required for Smad6 to interact with this GR fragment, whereas the N-terminal fragment GR-(263-319) is sufficient for supporting the interaction of G␤2, suggesting that Smad6 and G␤2 bind to different surfaces of the GR, although these proteins were found as GR-interacting molecules in the same yeast two-hybrid screening.
Smad6 Suppresses GR-induced Transcriptional Activity-We next examined the effects of overexpressed Smad6 on GR-induced transcriptional activity of the glucocorticoid-responsive MMTV promoter in HCT116 cells (Fig. 2). Increasing amounts of Smad6-expressing plasmid dose-dependently suppressed the transcriptional activity of the GR (Fig.  2A). Smad6 also suppressed GR transcriptional activity on the integrated MMTV promoter, which was introduced into the chromosome of HCT116/MMTV cells by a standard stable transfection procedure (Fig. 2B). Smad6 strongly suppressed the transcriptional activity of the wild type GR, whereas it did not influence that of GR-(⌬262-404), which is devoid of an interaction domain for Smad6 (Fig. 2C).
We also tested Smad6 on a well known endogenous glucocorticoidresponsive gene, the rat TAT. Glucocorticoids increase TAT enzymatic activity by stimulating the transcriptional rate of this gene via tandem GREs located in its promoter region (30). In rat HTC cells, 10 Ϫ6 M dexamethasone increased the TAT activity by 7-fold, whereas transfection of siRNAs for Smad6 significantly enhanced dexamethasone-stimulated TAT activity (Fig. 2D). The siRNA transfection reduced mRNA abundance of Smad6 in these cells (Fig. 2E). These results indicate that endogenous Smad6 acts as a negative regulator of GR transactivation on the endogenous glucocorticoid-responsive gene.
Smad6, but Not Smad7, Suppresses GR Transcriptional Activity via Its C-terminal MH2 Domain-To define a domain of Smad6, which is responsible for its suppression of GR transcriptional activity, we employed Smad7, several chimeras of Smad6 and Smad7, and their fragments, and tested them in the GR-induced transactivation of the MMTV promoter in HCT116 cells (Fig. 3, A and B). Although Smad7 shares high similarity with Smad6 and has several overlapping activities (14), Smad7 did not suppress GR transcriptional activity at all. Among tested chimeras and fragments, those harboring the C-terminal MH2 domain of Smad6, such as Smad6C and Smad7/6, preserved the wild type's suppressive effect on GR transactivation. Thus, the MH2 domain of Smad6 is critical for suppression of GR transactivation, possibly by supporting physical interaction with the GR. Although Smad7 has a similar MH2 domain, it has no effect on GR transactivation.
Smad6 Antagonizes Dexamethasone-stimulated Gluconeogenesis in Rat Liver by Suppressing PEPCK Gene Induction-We next examined the effect of Smad6 overexpression in the rat liver on circulating glucose levels and mRNA abundance of the PEPCK gene in order to verify Smad6-induced suppression of GR transactivation in vivo. We focused on glucocorticoid-induced gluconeogenesis, since it is a well known biological activity of glucocorticoids mediated by their transactivation property (5,31). PEPCK is a major rate-limiting enzyme of gluconeogenesis through which glucocorticoids stimulate glucose production in the liver (5). This organ is also known to express substantial amounts of Smad6 (32).
We injected the FLAG-Smad6-expressing adenovirus or the control adenovirus that expresses LacZ into the peritoneal cavity. After 24 h, we    DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51 injected 1 mg/kg of dexamethasone intramuscularly. After an additional 24 h, we measured blood glucose levels in whole blood and evaluated mRNA abundance of the PEPCK and TAT genes in the liver (Fig. 4). Dexamethasone increased blood glucose levels, whereas injection of FLAG-Smad6-expressing adenovirus suppressed this increase (Fig. 4A). In the liver, dexamethasone treatment increased PEPCK mRNA expression. Injection of the FLAG-Smad6-expressing adenovirus antagonized to this dexamethasone-induced increase of the PEPCK mRNA levels (Fig. 4B). mRNA expression of the TAT gene was also stimulated with the dexamethasone injection, and Smad6 adenovirus again suppressed the dexamethasone effect on this gene (Fig. 4C). In the whole homogenate of the liver, FLAG-Smad6 was detected in the rat liver injected with FLAG-Smad6-expressing adenovirus but not with the LacZ-expressing adenovirus (Fig. 4D). Levels of the GR protein were not altered throughout the experiment (Fig. 4D). These results indicated that Smad6 antagonizes dexamethasone-stimulated gluconeogenesis in the rat liver by suppressing the PEPCK gene induction, possibly by down-regulating GR transcriptional activity on this gene. We also obtained similar results in rats injected with the adenoviruses via the tail veins (data not shown).

Smad6 Is a Corepressor of GR
Smad6 Suppresses GR Transcriptional Activity by Attracting HDAC3-It is known that Smad6 and Smad7 suppress the transcriptional activity of the osteopontin promoter by binding to Hoxc-8-responsive elements as a dimer with Hoxc-8 and by attracting the class-I histone deacetylases, such as HDAC1 and -3 (17,18). The NH 2 domain of Smad6 interacts with these histone deacetylases (18). CtBP is also known to mediate the inhibitory effect of Smad6 on the Id1 promoter, but this protein does not interact with the MH2 domain of Smad6 (19). Therefore, we examined contribution of histone deacetylases in the Smad6-induced suppression of GR transactivation by using a histone deacetylase inhibitor TSA (Fig. 5A). In HCT116 cells transfected with the GR-expressing plasmid and pMMTV-Luc, the addition of the indicated concentrations of TSA weakly enhanced GR-induced transcriptional activity in the absence of Smad6. Smad6 expression strongly suppressed GR transactivation, and TSA significantly attenuated Smad6-induced suppression of the GR transcriptional activity in a dose-dependent fashion, with complete attenuation of the Smad6 effect at 100 nM concentration. These results indicate that Smad6 suppresses GR-induced transactivation of the MMTV promoter activity through histone deacetylation.
We next examined the effects of known HDACs on the GR-induced transactivation of the MMTV promoter in HCT116 cells (TABLE  ONE). In the absence of Smad6, overexpressed HDAC1, -3, -4, and -5, but not HDAC6, demonstrated significant suppressive effect on dexamethasone-stimulated GR transactivation. Among them, HDAC3 showed the strongest effect. Smad6 effectively suppressed GR transactivation, and only HDAC3, but not the other tested HDACs, further suppressed GR transactivation in a statistically significant fashion. These results may indicate that Smad6 suppresses GR transcriptional activity of the MMTV promoter by primarily cooperating with HDAC3.
We further examined association of GR with Smad6 and HDAC3 on the MMTV GREs in HCT116/MMTV cells by using ChIP assays (Fig. 5, B-D). FLAG-Smad6 was attracted to GREs with the wild type GR, whereas such attraction almost disappeared with GR-(⌬262-404) that lacks most of the Smad6 interaction domain (Fig. 5B). HA-tagged HDAC3 was successfully co-precipitated with GREs only when FLAG-Smad6 was transfected and attracted to GREs (Fig. 5C). Further, HA-HDAC3 was co-precipitated with GREs in the presence of the wild type FLAG-Smad6, but not in the presence of FLAG-Smad6N that does not have a C-terminal MH2 domain that interacts with GR (Fig. 5D).

Smad6 Antagonized p160 Histone Acetyltransferase Coactivator-induced Histone Acetylation and Enhancement of GR Transcriptional
Activity-GR stimulates the transcriptional activity by attracting transcriptional intermediate molecules, such as coactivators and cofactors, which modulate the chromatin structure and further communicate with the transcriptional machinery, including general transcription factors and the RNA polymerase II (8). Among them, histone acetyltransferase coactivators, such as p160 type nuclear receptor coactivators and p300/CREB-binding protein, plays an essential role in GR-induced transcriptional activation by acetylating specific lysine residues of the histone tail (33). Once histones are acetylated at their N-terminal tails, they allow accumulation of transcription factors, cofactors, and the RNA polymerase II on the promoter region (34). Since Smad6 suppresses GR transactivation by attracting HDAC3, we examined whether Smad6 antagonizes to dexamethasone-and p160 coactivator-induced acetylation of histones using ChIP assays with specific antibodies against acetylated histone H3 and H4 (Fig. 6, A-D). Histone H3 and H4 are well

. Adenovirus-mediated expression of Smad6 in intact rats suppresses dexamethasone-induced increase of circulating glucose levels (A) and mRNA abundance of the PEPCK (B)
and TAT (C) genes in the liver. Sprague-Dawley rats were intraperitoneally injected with 1 ϫ 10 10 colony-forming units of Smad6-expressing (ϩ) or LacZ-expressing (Ϫ) adenoviruses. They were then injected with 1 mg/kg dexamethasone intramuscularly and subsequently sacrificed. Glucose levels (A) were measured in whole blood, whereas mRNA abundance of the PEPCK (B), TAT (C), and control RPLP0 genes were determined in the liver by SYBR green-based real time PCR using their specific primer pairs. Expression levels of the FLAG-tagged Smad6 and the GR (D) were determined in Western blots with their specific antibodies in liver homogenate. *, p Ͻ 0.05; **, p Ͻ 0.01; n.s., not significant, compared with the base line.
known natural targets of the acetylation reaction promoted by p160 proteins (33). We used HTC cells, which express endogenous GR and the glucocorticoid-responsive TAT gene that has two GREs at its promoter region (30). Acetylated histone H3 and H4 appeared on the TAT GREs in response to dexamethasone addition, and overexpression of GRIP1 further potentiated their appearance. Expression of the wild type Smad6 reduced acetylation of H3 and H4 induced by dexamethasone and strongly attenuated GRIP1-potentiated accumulation of acetylated histones on GREs. Smad6N, which does not have the MH2 domain, lost all of these effects. In agreement with these results, Smad6 wild type, but not Smad6N, almost abolished GRIP1-induced enhancement of GR transcriptional activity on the MMTV promoter (Fig. 6E). These results  DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51 thus indicate that Smad6 facilitates deacetylation of histone H3 and H4, whose acetylation was induced by GR activation and subsequent attraction of histone acetyltransferases including p160 proteins.

Smad6 Suppresses the Transcriptional Activity of Several Steroid Hormone Receptors but Not That of the Estrogen Receptor and Has No Such
Effect on the Activities of p53, NF-B, and CREB-Since GR and other steroid hormone receptors share many characteristics in their structure, mechanisms of action, and use of coactivators and corepressors (8), we examined Smad6 on the transcriptional activity of other steroid hormone receptors (Fig. 7). We expressed steroid hormone receptors or other transcription factors with their responsive promoter-driven luciferase reporter genes in HCT116 cells (Fig. 7A). As expected, Smad6 suppressed mineralocorticoid receptor (MR)-, progesterone receptor-A (PR-A)-, and androgen receptor (AR)-induced transcription of their responsive genes in addition to that of GR. Smad6, however, did not affect transcriptional activity of ER␣, p53, NF-B, and CREB. Consistent with these functional analyses, N-terminal domains of MR, PR-A, and AR interacted with Smad6 (331-496) in a yeast two-hybrid assay similar to the GR, whereas that of ER␣ failed to associate with this portion of Smad6 (Fig. 7B).

DISCUSSION
Smad6 interacted with GR at amino acids 263-419, located between the AF-1 and DBD of the GR, using its C-terminal MH2 domain for this interaction. Smad6 suppressed GR-induced transcriptional activity of glucocorticoid-responsive genes, such as the transiently transfected and integrated MMTV promoter and the endogenous glucocorticoid-responsive TAT gene. Overexpressed Smad6 in the rat liver inhibited glucocorticoid-induced increase of blood glucose levels and induction of PEPCK gene expression, suggesting that Smad6 antagonizes glucocorticoid-induced gluconeogenesis in the liver in vivo. Although Smad7 is highly homologous to Smad6, it had no effect on GR-induced transcriptional activity.  Although definitely less ubiquitous than the GR, Smad6 is widely expressed in the organism. Smad6 mRNA is normally found in the lung, kidney, liver, heart, and, in small amounts, also in the brain and the skeletal muscle (32). It is thus possible that Smad6 regulates glucocorticoid activity in these organs in addition to the liver, where it suppresses glucocorticoid-stimulated gluconeogenesis. Since TGF␤, activin, and BMP-7 regulate the expression levels of Smad6 (20), it is likely that these TGF␤ family members indirectly regulate glucocorticoid action in target tissues through Smad6. To emphasize the potential implications of our findings, the complex bidirectional interactions between the signaling systems of glucocorticoids and TGF␤ family proteins are summarized in TABLES TWO and THREE. These mutual influences suggest that the antiglucocorticoid properties of Smad6 may be of major importance in maintaining some of the actions of glucocorticoids under restraint. The Smad6 pathway in the cell may thus contribute to the antineurotoxic, anticatabolic, and pro-wound healing properties of TGF␤ family members, which oppose these negative actions of glucocorticoids.
Smad6 knock-out mice develop multiple cardiovascular abnormalities, such as those caused by endocardial cushion transformation and aortic ossification and have elevated blood pressure in adulthood (35). We suggest that alterations of glucocorticoid actions in target organs may also be developed in Smad6 KO mice. Indeed, organ-or tissuespecific alterations of glucocorticoid sensitivity may produce diverse pathologies, including highly prevalent disorders, such as the metabolic syndrome and depression, that account for a major proportion of human morbidity and mortality (36,37).
In addition to GR, Smad6 also suppressed the transcriptional activities of several other steroid hormone receptors, including the MR, PR-A, and AR (but not that of ER␣ or of the transcription factors p53, NF-B, and CREB), possibly via direct interaction with their N-terminal, immunogenic domains. MR, AR, and PR-A are phylogenetically closer to GR than ER␣, with relatively long N-terminal domains (38). Thus, these steroid hormone receptors quite likely interact with Smad6 via their N-terminal domains, employing it as an intracellular regulator of their ligand-induced activity. Smad6 hence appears to be a general corepressor of several steroid hormone receptors, negatively regulating the action of these steroids in their target organs in response to TGF␤ family members.
Smads other than Smad6 interact with members of nonsteroidal nuclear receptor family proteins. Thus, Smad3 interacts with the vitamin D receptor and enhances its transcriptional activity together with p160 coactivators (39,40). Smad7, but not Smad6, suppresses this Smad3-induced enhancement of VDR activity (40). Smad3 also interacts with the GR and the AR, and activation of these receptors sup-  Smad6 Is a Corepressor of GR DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51 presses Smad3-induced transactivation (41,42). Agonists of peroxisome proliferator-activated receptor-␣ inhibit TGF␤-responsive ␤5 integrin expression and disrupt formation of the Sp1-Smad4 complex, in parallel with induction of peroxisome proliferator-activated receptor-␣/Smad4 interaction (43). We also found that dexamethasone moderately suppressed Smad6-induced suppression of Smad5 transactivation of a BMP response element-driven promoter (data not shown). These pieces of evidence indicate that there are numerous cross-talks between the Smad-and nuclear receptor-mediated signaling systems, consistent with the biologic importance of these discrete pathways.
The inhibitory effect of Smad6 on GR transcriptional activity was mediated by the histone deacetylase HDAC3. Indeed, HDAC3 was attracted to GREs through the GR-Smad6 complex and antagonized GRIP1-induced acetylation of histones H3 and H4. Classic corepressors, such as the nuclear receptor corepressors and the silencing mediator for retinoid and thyroid hormone receptor, suppress transcriptional activity of unliganded nuclear receptors, such as the retinoic acid and thyroid hormone receptors, by attracting histone deacetylases, contributing to silencing of their responsive genes in the absence of ligands (8). Unliganded GR does not bind these corepressors, but glucocorticoid antagonists like RU 486 induce interaction of antagonist-bound GR with the nuclear receptor corepressors and the silencing mediator for retinoid and thyroid hormone receptor, partially explaining the inhibitory effect of these molecules (44 -46). In contrast to these classic corepressors, Smad6 and HDAC3 are attracted to agonist-activated GR and suppress GR-induced transactivation. Smad6 interacts with the N-terminal domain of the GR, whereas classic corepressors interact with the ligandbinding domain as well as the AF-1 domain (44,45). The attracted Smad6-HDAC3 complex then antagonizes p160 histone acetyltransferase-induced acetylation of H3 and H4 and suppresses GR-induced transactivation (Fig. 8).
Smad6 may play a physiologic role in the termination of GR transactivation initiated by the attraction of histone acetyltransferase coactivators, allowing GR to restart a new cycle of transcriptional activation (10 -12), in contrast to classic corepressors, which play a role in the silencing of genes in the absence of ligand or presence of antagonists (8,(45)(46)(47). It is also possible that Smad6/HDAC3 deacetylate other GRattracted transcriptional components, such as the p160 or p300/CBP histone acetyltransferases and possibly the GR itself. Through this mechanism, Smad6 might further regulate ligand-bound GR-induced transcriptional activity. Acetylation of these molecules regulates their influence on transcription, whereas Smad7 deacetylates itself via attracting HDACs, further regulating its own stability and activity (48 -50).